{ "google/gemma-2b": { "backend.cache_implementation=null,backend.torch_compile=False": { "2e3f8f74747deeeead6cf1f0c12cf01bd7169b82": { "metrics": { "prefill.latency.mean": 0.020818703651428225, "prefill.throughput.value": 336.2361133143744, "decode.latency.mean": 2.5248167724609374, "decode.throughput.value": 50.30067979000836, "per_token.latency.mean": 0.019879149895014724, "per_token.throughput.value": 50.303961954166816 } } }, "backend.cache_implementation=static,backend.torch_compile=False": { "2e3f8f74747deeeead6cf1f0c12cf01bd7169b82": { "metrics": { "prefill.latency.mean": 0.02216444778442383, "prefill.throughput.value": 315.8210873595183, "decode.latency.mean": 2.5914827880859375, "decode.throughput.value": 49.00669245571254, "per_token.latency.mean": 0.020404336876756562, "per_token.throughput.value": 49.00918888175886 } } }, "backend.cache_implementation=static,backend.torch_compile=True": { "2e3f8f74747deeeead6cf1f0c12cf01bd7169b82": { "metrics": { "prefill.latency.mean": 0.014170464038848876, "prefill.throughput.value": 493.98523441499367, "decode.latency.mean": 1.567123046875, "decode.throughput.value": 81.04022224244018, "per_token.latency.mean": 0.012338381572032537, "per_token.throughput.value": 81.04790682326639 } } } } }